Agricultural and Forest Meteorology 198–199 (2014) 94–104

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology

journal homepage: www.elsevier.com/locate/agrformet

Applicability and limitations of using the crop water stress index as an

indicator of water deficits in citrus orchards
V. Gonzalez-Dugo a,∗ , P.J. Zarco-Tejada a , E. Fereres a,b
a
Instituto de Agricultura Sostenible (IAS), Consejo Superior de Investigaciones Científicas (CSIC), Alameda del Obispo s/n, 14004 Cordoba, Spain
b
Department of Agronomy, University of Cordoba, Campus Universitario de Rabanales, 14014 Cordoba, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Accurate measurements of the crop water status are becoming essential in irrigated agriculture, as water
Received 23 January 2014 resources are limited and its use must be optimized, especially in semi-arid conditions. Indicators derived
Received in revised form 4 August 2014 from thermal information have shown to be closely related to water status in several fruit tree species,
Accepted 5 August 2014
and have shown promise for assessing the spatial variations among and within whole orchards. Here, a
Available online 29 August 2014
methodology is proposed for assessing the Crop Water Stress Index (CWSI) of mandarin (Citrus reticulata
Blanco cv. Clemenvilla) and navel orange (Citrus sinensis L cv. Powell) located in Southern Spain, taking into
Keywords:
account the short-term fluctuations in canopy temperature observed in both species. Infrared thermal
CWSI
Irrigation scheduling sensors were installed above trees to record canopy temperature (Tc ) continuously for three seasons
Deficit irrigation from 2009 to 2011. The Non Water Stress Baseline (NWSB) was calculated using Tc of well irrigated
Thermal imagery trees on cloudless days during summertime. The NWSB was affected by flushes of growth that occurred
UAV periodically, depending on the crop load of a given year. Nevertheless, the close relationship observed
between CWSI and stem water potential (R2 ranging between 0.59 and 0.66; p < 0.001) demonstrated
that it is a suitable indicator of water status in citrus. The results showed that care must be taken when
using CWSI in citrus to account for the presence of new growth at the top of the canopy, and for the
short-term fluctuations in canopy temperature. The canopy temperature information acquired from point
sensors was used in conjunction with high resolution airborne thermal imagery to derive CWSI maps. The
approach presented here demonstrates that CWSI is a valuable method to assess the water status, and to
quantify the spatial variability in water stress among and within citrus orchards using high-resolution
airborne thermal imagery.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction
It has long been known that canopy temperature and tran-
spiration are closely related (Gates, 1964). Water stress induces
stomatal closure, reduces evaporative cooling and increases leaf
temperature (Hsiao, 1973). This feature was used by Jackson et al.
(1977, 1981) in the development of water stress indicators based on
canopy temperature. Working with hand-held infrared thermome-
ters, Jackson and coworkers developed the concept of Crop Water
Stress Index (CWSI; Jackson et al., 1981; Idso et al., 1978). This indi-
cator is based on normalizing the difference between canopy (Tc )
and air temperature (Ta ) relative to the evaporative demand (by
means of the vapor pressure deficit of the air, VPD). The normal-
ization related to VPD considers the Tc –Ta difference of a canopy
∗ Corresponding author. Tel.: +34 957499170; fax: +34 957499252.
E-mail address: victoria.gonzalez@ias.csic.es (V. Gonzalez-Dugo).
under two boundary conditions: (a) a lower limit when it transpires
at its potential rate (i.e., under well watered conditions) and (b) an
upper limit under no transpiration. The lower limit is described by
a linear regression between Tc –Ta and VPD, which is known as the
Non Water Stress Baseline (NWSB). There are several methods to
calculate the upper limit (corresponding to the Tc –Ta of a canopy
where transpiration is halted). The most widely used was proposed
by Idso et al. (1981) and defines the upper limit as a horizontal line
departing at the intercept of the NWSB corrected for air temper-
ature. These two bounds are climate dependent and are used to
calculate the CWSI when plotted as a function of VPD (Idso et al.,
1981). Analytical solutions to assess the CWSI have also been devel-
oped (Jackson et al., 1981; Jones, 1999; Maes and Steppe, 2012)
that enabled the calculation of CWSI at the canopy level, which is
of paramount importance for irrigation management applications.
The CWSI has been extensively used in annual crops (Idso et al.,
1981; Wanjura et al., 1990; Irmak et al., 2000; González-Dugo et al.,
2006; Kar and Kumar, 2007). Recently, it has also been applied to

http://dx.doi.org/10.1016/j.agrformet.2014.08.003
0168-1923/© 2014 Elsevier B.V. All rights reserved.
 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104 95

perennial crops, such as olive (Ben-Gal et al., 2009; Berni et al., and orange trees. A methodology to quantify the CWSI was estab-
2009), pistachio (Testi et al., 2008) and peach (Wang and Gartung, lished and tested over three seasons, and the extrapolation from
2010). thermal point data to thermal imagery, remotely acquired by an
Judicious irrigation management is based on a precise estimate airborne platform was carried out.
of the crop water requirements. The traditional use of reference
evapotranspiration (ETo) and crop coefficients (Kc ) proposed by
2. Materials and methods
Allen et al. (1998) is now being substituted, in the case of tree
crops, by more accurate models to compute water use by trees
2.1. Site description and experimental design
(Rana et al., 2005; Fereres et al., 2012; Villalobos et al., 2013). In the
case of citrus, maximum water requirements may be assessed with
The study was carried out during three years (2009–2011) in
a simple model as function of ground cover (Villalobos et al., 2009).
two 0.6-ha plots of orange and mandarin trees in a commercial
Often, the full water requirements cannot be satisfied and thus,
orchard located near La Campana, Seville (37.8◦ N, 5.4◦ W), Spain.
deficit irrigation strategies are being imposed as water resource
The orange trees (Citrus sinensis L. cv. Powell) were planted in
becomes increasingly limited. Deficit irrigation is also proposed
1997 in a 7 m × 4 m grid (358 trees/ha) on a deep alluvial soil of
as a strategy for increasing fruit quality (Ballester et al., 2011).
loam to sandy-loam texture. The mandarin (Citrus reticulata Blanco
Growers must decide how to use the water efficiently in order
cv. Clemenvilla) orchard was planted in 1997 in a 7 m × 3 m grid
to maintain yields and enhance fruit quality. Accurate water sta-
(476 trees/ha). Both cultivars were grafted onto Carrizo citrange
tus assessments become essential in order to manage irrigation
(C. sinensis L. Osb. × Poncirus trifoliata L. Osb), and were around
in water scarcity situations (Naor, 2006), and the search for reli-
3 m height. The climate in the area is Mediterranean, character-
able indicators of tree water status has recently expanded at the
ized by warm and dry summers and cool and wet winters, with
individual tree (Fernandez and Cuevas, 2010) and at the field level
an average annual rainfall and reference evapotranspiration (ETo;
(Bastiaanssen and Bos, 1999; Gonzalez-Dugo et al., 2013).
Penman-Monteith) around 550 and 1300 mm, respectively.
One of the main advantages of the thermal derived indicators is
Four treatments were compared in a randomized block design
that they can be remotely acquired and thus adapted to operate at
with four replicates per treatment. The experimental design was
the field and farm scales by remote sensing. Point indicators, which
similar for orange and mandarin. Each individual plot comprised 12
are currently the state of the art, typically assess water status at
trees (three rows and four trees in each row), the two central trees
the leaf level and extrapolate those values at the plant level. How-
being monitored. Trees were daily irrigated with an automated
ever, irrigation management (and especially when deficit irrigation
drip system, with ten pressure-compensated emitters (1.6 l h−1 )
strategies may be applied) requires the precise assessment of water
per tree. The treatments were: (i) C, where water was supplied to
status at the field scale, including within-field variability. Further-
satisfy the full water requirements throughout the season, (ii) RDI1,
more, field characterization by point indicators is time consuming
where water supply was reduced to 37% of C from mid-June to early
and does not represent well field heterogeneity. Remotely-sensed
September, (iii) RDI2, with the same schedule as RDI1, but with a
derived indicators allow the characterization of an entire field while
milder water deficit during that period, established at 50% of C; and,
maintaining high resolution to enable a close relationship with
(iv) F, the commercial irrigation management followed in the rest
individual plant performance and the underlying physiology (Berni
of the orchard (based on long-term average ETo and locally devel-
et al., 2009; Gonzalez-Dugo et al., 2012; Zarco-Tejada et al., 2012).
oped crop coefficients that includes an adjustment in canopy size).
An issue that requires investigation is whether the thermal indica-
In the deficit treatments, irrigation was modified by using lower
tors obtained from ground sensors in citrus can be extrapolated to
discharge emitters while maintaining the number of emitters and
airborne or satellite thermal imagery, as previously proposed for
the application time. The starting date for the two RDI treatments
olive (Sepulcre-Cantó et al., 2006).
varied among years, and occurred around the end of the physiolog-
World citrus production has increased in recent decades
ical June fruit drop. Full water requirements were estimated based
(Spreen, 2001) and is still increasing (FAOSTAT, 2012). Nowadays,
on the crop evapotranspiration, which was calculated following
the Mediterranean region accounts for more than 16% of the total
Goldhamer, 2012. ETo was obtained from an automated weather
world production of citrus equivalent to about one million hectares
station located 10 km away from the study site. The farm min-
of land use (FAOSTAT, 2012). These data highlight the importance
eral fertilization program included periodic applications through
of citrus production in the region. Given the semi-arid climate of
the irrigation system of about 220 kg/ha of N, 90 kg/ha of P2 O5 ,
the Mediterranean basin, irrigation is essential for citrus produc-
and 110 kg/ha K2 O. To compensate for the differences in irrigation
tion. Traditionally, citrus has been irrigated by flooding, although
volumes applied in the RDI treatments, additional fertilizer was
water scarcity and management reasons have led to a shift to drip
applied through the system to achieve the same level of fertility in
irrigation in many production areas. Investments in irrigation hard-
all treatments.
ware should be accompanied by more precise management of the
scarce resource. Gonzalez-Dugo et al. (2013) have used the concept
of CWSI to assess its potential for irrigation management in a com- 2.2. Canopy temperature measurement and CWSI calculation
mercial orchard with several tree crops. In the case of citrus, to our
knowledge, there is only one published report on the use of CWSI in Eight trees (four orange and four mandarin trees) were instru-
sweet lime (Sepaskhah and Kashefipour, 1994), which has a differ- mented with eight infrared temperature (IRT) sensors with an
ent pattern of water use from other citrus species (Sepaskhah and angular field of view of 44◦ (Apogee, Utah, USA) acquiring continu-
Kashefipour, 1995). Recent work with orange trees in the coastal ous thermal crown temperature data from June 2009 to September
area of Spain (Ballester et al., 2013a) was unable to characterize the 2011 in the mandarin and orange orchards. Each sensor was
CWSI under contrasting water regimes. Given the potential use- installed 1 m over the targeted tree crown, on aluminum masts with
fulness of the CWSI for remote sensing applications (Berni et al., a horizontal arm that ended up over the center of the canopy of two
2009; Maes and Steppe, 2012), it would be important to establish consecutive trees, being both monitored with one IRT sensors each.
if the CWSI can be quantified in citrus plantations, and what are its For each species, two trees of C treatment and two trees of RDI1
limitations that have prevented its determination until now. were monitored, and their canopy temperatures were averaged
This work was aimed at testing the hypothesis that the Crop to obtain an hourly value per species and treatment. For addi-
Water Stress Index is a reliable water stress indicator in mandarin tional information on sensor installation, see Berni et al. (2009). We
96 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104
consider that only pure tree crown temperatures were monitored
by the sensors, given that a typical citrus tree has over 140 m2 of leaf
area (Turrell, 1961) and that is equivalent to a leaf area index of over
13 under the orchard spacing and canopy size of our experimen-
tal trees. The accuracy of the sensors yielded ±0.15 ◦ C, according
to the manufacturer. These sensors were connected to a datalog-
ger (model CR1000, Campbell Scientific, Logan, UT) that recorded
canopy temperature every minute and stored the 5-min averages.
For each species, two trees from C and RDI1 treatments were mon-
itored. During 2010, problems with the dataloggers disabled the
registration of all the data in two periods in the mandarin, (from
DOY 181 to 197 and from 227 to 235), and from DOY 174 to 202 in
the orange.
Air temperature (Ta , K) and relative humidity (RH, %) were
recorded continuously with a Vaisala Weather Transmitter (model
WXT510, Vaisala Oyj, Helsinki, Finland) installed in the study site
1 m over the tree crowns in the mandarin experiment.
Canopy temperature (Tc ) and air temperature (Ta ) were used to
calculate the Crop Water Stress Index (CWSI), according to Eq. (1):
(Tc − Ta ) − (Tc − Ta )LL
CWSI =
(Tc − Ta )UL − (Tc − Ta )LL
Where (Tc –Ta )LL is the lower limit of the difference between
canopy and air temperature. It corresponds with the Tc –Ta value
of a canopy transpiring at the potential rate. (Tc –Ta )UL corresponds
with the difference of a canopy where transpiration is completely
halted.
The relationship between Tc –Ta and VPD for well irrigated trees
was used to establish the Non-Water-Stressed Baseline (NWSB)
which defines the lower limit ((Tc –Ta )LL ) for a given evaporative
demand. (Tc –Ta )UL was calculated as the intercept of the NWSB cor-
rected for air temperature, according to the methodology described
by Idso et al. (1981). From the entire seasons, only cloudless days
from mid June to end of August were selected for the calculation of
the CWSI.
NWSB was calculated for Tc –Ta values at solar noon. In man-
darin in 2011, the period comprised between DOY 176 and 196
was discarded due to a failure in the irrigation system that altered
the anticipated treatments.
2.3. Aerial imagery
An unmanned aerial vehicle (UAV) platform for remote sensing
research was developed at the Laboratory for Research Methods in
Quantitative Remote Sensing (QuantaLab, IAS-CSIC, Spain) to carry a
payload with a thermal and a hyperspectral camera. Two UAV plat-
forms were operated in this experiment. The first UAV consisted of a
2-m wingspan fixed-wing platform capable of one-hour endurance
at 5.8 kg take-off weight (TOW) (mX-SIGHT, UAV Services and Sys-
tems, Germany). This platform was used to fly the thermal camera
over the study sites in 2009. A second UAV platform was devel-
oped for hyperspectral imagery acquisition, consisting of a 5-m
wingspan fixed-wing platform capable of three-hour endurance
at 13.5 kg TOW (Viewer, ELIMCO, Seville, Spain). This larger plat-
form enabled the image acquisition of both study sites in a single
flight carrying both the micro-hyperspectral imager and the ther-
mal camera concurrently. Platform characteristics are detailed in
Zarco-Tejada et al. (2012).
On DOY 217 2009, images of the experimental plots were
acquired with a thermal camera (MIRICLE 307, Thermoteknix
Systems Ldt., Cambridge, UK) installed onboard the UAV. Flight
altitude was 250 m above the ground, and spatial resolution was
35 cm (pixel size). Images were radiometrically and atmospheri-
cally calibrated as described in Zarco-Tejada et al. (2012). Canopy
temperature (Tc ) was extracted from the individual images for each
tree of the experiment. An algorithm was applied to restrict the
shape of the regions of interest considered and to exclude crown
edges, thus avoiding soil/vegetation mixed pixels. Once the Tc of
pure crowns was obtained, CWSI at the object level was calculated
after Eq. (1).
A micro-hyperspectral camera (Micro-Hyperspec VNIR model,
Headwall, Photonics, MA, USA) was flown over the orange experi-
ment on DOY 257 2010 and DOY 263 2011. Information concerning
radiometric calibration of the imager and atmospheric correction
of the imagery can be found in Zarco-Tejada et al. (2012). Flight
altitude was 550 m above the ground, yielding 35 cm resolution in
260 bands in the 400–900 nm spectral range. Similarly to what was
described for the thermal imagery, pure crowns were targeted to
avoid background effects on the spectra. The chlorophyll content
of the tree crowns was estimated based on the data extracted from
the hyperspectral images and the model developed by Zarco-Tejada
et al. (2004) for open-canopy tree crops.
2.4. Tree measurements
(1) Midday stem water potential (SWP) was monitored weekly in
eight trees per treatment and per species throughout the seasons
using a pressure chamber (Soil Moisture Equipment Corp. model
3000, Santa Barbara, CA, USA). Two shaded leaves located near the
trunk per tree were used to obtain an estimate of SWP.
Crop load (#fruits tree−1 ) and yield (kg tree−1 ) were determined
at harvest separately for each monitored tree.
In order to assess the spectral response of crowns displaying
contrasted vegetation characteristics, a series of leaf-level mea-
surements was made, including SPAD values obtained with a
chlorophyll meter (SPAD-502, Minolta Corp., Ramsey, NJ) on ten
sun-exposed fully-expanded young leaves recently developed from
the current summer growth, and on ten over-wintered adult leaves
(leaves that had appeared, at least, six months before), all sun-
exposed. SPAD values were used to determine chlorophyll content
(Cab) according to the model developed by Jifon et al. (2005). The
assimilation (A) and transpiration rate (Tr) at the leaf level, as well
as leaf temperature were also determined in these leaves with
a portable photosynthesis measurement system (LCpro-SD, ADC
BioScientific Ltd., Herts, England). Measurements were performed
around noon.
3. Results
3.1. Diurnal evolution of Tc –Ta in citrus and temperature
oscillations
Fig. 1 illustrates the patterns of the daily evolution of Tc for
mandarin in control and stressed trees, as well as the Tc dif-
ferences between both treatments, before the onset of stress
(Fig. 1A) and with increasing levels of water stress (Fig. 1B
and C). We noted that Tc experienced large oscillations, espe-
cially around midday, with a variable period ranging between
20 and 60 min. Within these oscillations, Tc often varied more
than 1 ◦ C. On DOY 162 (Fig. 1A), trees in both treatments showed
no Tc differences, reaching a maximum value of about 41.5 ◦ C.
In subsequent days, the difference between control and stressed
trees became significant (Fig. 1B and C) and increased with
time. Differences between treatments attained a value as high
as 3.5 ◦ C. The vapor pressure deficit (kPa) and incoming solar
radiation (W m−2 ) for the selected days did not explain these fluc-
tuations in Tc (Fig. 1D–F). It can also be observed that, as the stress
level increased, the Tc differences between treatments started ear-
lier in the day. Differences between both treatments generally
disappeared at around 17:00 (GMT), as a result of the decline in the
 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104 97

45 7 7 1200
TcC A VPD D
6
40 TcRDI 6 Rs 1000
5
TcRDI-TcC 5
4
35 800

TcRDI-TcC (ºC)
3

Rs (W/m2)
VPD (kPa)
4
Tc (ºC)
30 2 600
3
1
25 400
0 2
-1
20 1 200
-2
15 -3 0 0
0:00 6:00 12:00 18:00 0:00 0:00 6:00 12:00 18:00 0:00
GMT GMT

45 7 7 1200
Tc C B 6
VPD E
40 6 1000
Tc RDI 5 Rs
TcRDI-TcC 4 5
35 800

TcRDI-TcC (ºC)
3

Rs (W/m2)
VPD (kPa)
4
Tc (ºC)

30 2 600
1 3
25 400
0 2
-1
20 1 200
-2
15 -3 0 0
0:00 6:00 12 :00 18 :00 0:00 0:00 6:00 12 :00 18 :00 0:00
GMT GMT
45 7 7 1200
TcC VPD F
C 6
40 TcRDI 6 Rs 1000
5
TcRDI-TcC
4 5
35 800
TcRDI-TcC (ºC)

Rs (W/m2)
4
VPD (kPa)
Tc (ºC)

30 2 600
1 3
25 400
0
2
-1
20 200
-2 1

15 -3
0 0
0:00 6:00 12 :00 18 :00 0:00
0:00 6:00 12 :00 18 :00 0:00
GMT
GMT

Fig. 1. Representative diurnal course patterns of canopy temperature, Tc (◦ C), and of its difference between treatments, for C and RDI1 in mandarin trees (left column; plots
A–C) and evolution of vapor pressure deficit (kPa) and incoming solar radiation (W m−2 ) (right column; plots D–F) on DOY 162 (A and D), DOY 199 (B and E) and DOY 234 (C
and F), in 2009. For figures A–C, each reading corresponded to the averaged value from two trees of the same treatment.

evaporative demand and in the incoming solar radiation (Fig. 1D–F).

The orange trees displayed similar results (data not shown).
3.5

3.2. Establishment of the NWSB 3

The ample, short-term fluctuations observed in Tc made the 2.5

computation of the relationship between Tc –Ta and VPD difficult.
Tc-Ta (°C)

2
We averaged the Tc values over several time periods to define accu-
rately the optimum time interval for the establishment of the NWSB 1.5
in these two cultivars. We tested different time intervals, ranging
30min y = -0.42 x + 3.76 R2 =0.49
from 10 to 120 min (every 10 min) and centered at noon. Although 1
the regression lines obtained for each of the intervals considered 60min y = -0.43x + 3.64 R2 =0.55
did not change by much, the best adjustment was observed for 60- 0.5
2
min Tc averages (data not shown). Fig. 2 showed some examples of 120min y = -0.44 x + 3.72 R =0.51
0
the regressions obtained for the mandarin site in 2009. Thus, for
0 2 4 6
all the subsequent analyses, 60-min Tc –Ta and VPD averages were
VPD (kPa)
used for the CWSI calculations. Tests performed in the remaining
datasets of both cultivars showed similar results. Fig. 2. Relationship between Tc –Ta (◦ C) and VPD (kPa), using 5 min observations
The Non Water Stress Baseline (NWSB) was established using averaged over 30, 60, and 120 min (centered at noon) obtained in two well-watered
measurements of Tc –Ta of control trees in cloudless days, averaged mandarin trees in 2009.
98 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104

6 6
A B
5 5

4 4

Tc-Ta (ºC)

Tc-Ta (ºC)
3 3

2 2009 2 2009

1 2010 1 2010
2011 2011
0 0
0 2 4 6 0 2 4 6
DPV (kPa) DPV (kPa)

Fig. 3. Relationship between Tc –Ta (◦ C) and VPD (kPa) for well-watered mandarin (A) and orange (B) trees on clear sunny days. Measurements from mid June until the end
of August were averaged between 11:30 and 12:30 (GMT). Regression lines for the dataset pooling 2009 and 2011 are shown in the graphs.

between 11:30 and 12:30 GMT, and the plots of Tc –Ta vs. VPD gen-
erated the NWSB lines (Fig. 3). This relationship was significant for
both mandarin (Fig. 3A) and orange (Fig. 3B) trees, with moder-
ate R2 values, ranging between 0.33 and 0.56 (p < 0.001; Table 1).
It can be observed that, although the slope of the regressions was
more or less the same for each species during the three years, there
were differences in the intercept among years (Fig. 3). The inter-
cept varied from 3.3 ◦ C to 3.9 ◦ C in mandarin and from 3.6 ◦ C to
4.5 ◦ C in orange. The minimum value of the intercept was found in
2010 in both species; for the same range of VPD, Tc –Ta was about
1 ◦ C lower as compared to the two other years. Data from 2009 and
2011 were pooled together, as the linear regression corresponding
to both years was nearly identical to those computed for separate
years (Table 1).
The moderate R2 values of the NWSB may be attributed to the
described fluctuations in canopy temperature. The examination of
data obtained in different days displaying close values of Ta and
VPD showed that such temperature variations were causing large
part of the variability observed in Fig. 3. It can be thus expected
that the deviation should be symmetrically distributed around a
mean value, as, for the same VPD, Tc would vary above or below its
mean value. Consequently, negative values of CWSI (corresponding
to those observations lying below the NWSB; see Fig. 3) should be
expected for control trees.
3.3. Variability among years
The differences observed in the intercept of the NWSB among
years were analyzed. In contrast to the two other years, fruit yields
in 2010 were very low (27 and 8 kg tree−1 for the targeted con-
trol trees in mandarin and orange, respectively). Yields in 2009
and 2011 averaged 64 and 82 kg tree−1 for mandarin and orange,
respectively. If the variation in yield among years had an effect over
the vegetation of the monitored tree areas located on the top of the
canopy, it could influence their heat dissipation characteristics and
Table 1
Parameters of the NWSB (Tc-Ta=a·VPD + b) for orange and mandarin and for each
experimental year, obtained between 13:30 and 12:30 GMT. The regression line for
2009 and 2011 data is also shown.
thus, the temperature registered by the IRT sensors. We observed
active flushes of new growth during the summer of 2010, the
low yield year, mainly in the upper parts of the canopy. Newly
formed leaves display a higher stomatal conductance, less pig-
ment contents, and thus different water relations than mature
leaves. These differences were assessed here with a series of mea-
surements made at the leaf level, where gas exchange rate and
chlorophyll content of ten young and ten over-wintered leaves
were measured. Table 2 shows that young leaves displayed a higher
transpiration rate than mature leaves (about 280% more), leading to
a decrease of about 1 ◦ C in mean leaf temperature. A lower value of
chlorophyll content in the young leaves (28 ␮g cm−2 ) was observed
relative to that of mature leaves (Table 2).
The differences observed in the chlorophyll content at the
leaf level were used to depict a change in vegetation pattern at
the top of the canopy by the hyperspectral imagery remotely
acquired over the orchard. The high resolution of the hyperspec-
tral imagery enabled the analysis of pure-crown spectra, avoiding
background effects (Fig. 4A). Crown spectra of the monitored trees
were extracted for each tree of the experiment from the images
acquired in 2010 and 2011. Fig. 4B shows the differences observed
in the crown reflectance for the control trees between the two
years. The chlorophyll content (Cab) calculated according to Zarco-
Tejada et al. (2004) showed that the tree crown had a lower amount
of Cab in 2010, the low yield year, as compared to 2011 (55.8 vs.
63.1 ␮g cm−2 ). These differences in Cab may be related to an accel-
erated development of young leaves that displayed a lower content
of Cab as compared to the mature leaves (Table 2). These young
leaves in 2010 altered the exchange rate and heat dissipation rela-
tive to the other two years. Chlorophyll content was also estimated
for the stressed trees but the mean chlorophyll values were nearly
identical for 2010 and 2011 (64.9 and 65.9 ␮g cm−2 ).
If the difference in the intercept of NWSB between 2010 and the
other two years (Fig. 3) would be related to the increased vegeta-
tive growth, presumably activated by the low amount of assimilates
demanded by the limited fruit number in 2010, one would expect
a positive relationship between crop load and the intercept of the
NWSB. Fig. 5 presents the relationship between crop load and the
intercept of the NWSB for each year and the two species. The higher

Species Year a b R2 P
Table 2
Mandarin 2009 –0.430 3.640 0.54 <0.001 Leaf temperature, gas exchange rate and chlorophyll content measured on ten young
2010 –0.496 3.313 0.52 <0.001 and mature (over-wintered) orange leaves. Leaf T: Leaf temperature, E: leaf transpi-
2011 –0.401 3.903 0.46 <0.001 ration, Gs: stomatal conductance, A: assimilation rate, Cab: chlorophyll content.
2009&2011 –0.500 4.056 0.56 <0.001
Leaf Leaf T E Gs A Cab
Orange 2009 –0.372 4.462 0.39 <0.001
type (◦ C) (mol m–2 s–1 ) (mmol m–2 s–1 ) (mmol m–2 s–1 ) (␮g cm–2 )
2010 –0.433 3.635 0.45 <0.001
2011 –0.320 4.487 0.33 <0.001 Young 38.7 4.33 126 7.42 28
2009&2011 –0.379 4.586 0.41 <0.001 Mature 39.7 1.53 28 3.02 61
 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104
Fig. 4. (A) Detail of the imagery acquired with the hyperspectral camera, showing the identification of targeted tree crowns. (B) Tree crown reflectance extracted from the
same well watered orange tree in 2010 and 2011, showing spectral differences between the two years.
5.0
4.8
4.6
NWSB intercept (ºC)
4.4
4.2
4.0
3.8
3.6
3.4 Mand arin
3.2 Orange
3.0
0 50 100 150
Crop load (fruits/tree)
Fig. 5. Relationship between the NWSB intercept and the crop yield (fruit tree−1 )
for orange and mandarin well-watered trees. Each point represents one year and
corresponds to the averaged value of the two monitored trees.
the crop load, the higher the intercept, each value presumably cor-
responding to a lower level of vegetative growth. Even though there
are only six data points in Fig. 5, the trend shown is consistent with
the proposed hypothesis.
3.4. Seasonal evolution of Tc –Ta
The seasonal course of Tc –Ta showed differences between C and
stressed (RDI1) trees, being these differences generally greater as
the stress level increased during the season. Fig. 6 shows the time
course of Tc –Ta measured at noon for mandarin and orange trees
in 2009 (Fig. 6A and D), 2010 (Fig. 6B and E) and 2011 (Fig. 6C
and F). Except for mandarin in 2011 (Fig. 6B), control trees main-
tained rather stable values, while Tc –Ta in stressed trees increased
during the experiment. Maximum differences between control and
stressed trees ranged between 1.5 and 2.5 ◦ C. In 2011, stressed trees
were rewatered following the achievement of severe water stress
levels. Recovery was performed on DOY 180 for the orange plots
and in DOY 190 for the mandarin plots. Once the water status of
the trees was recovered to normal levels, the irrigation treatments
resumed.
The relationship between Tc –Ta and the SWP measured both at
noon showed a correlation when data were pooled together, with
an R2 of 0.56 and 0.55 for mandarin and orange, respectively (Fig. 7).
99
3.5. Assessment of CWSI
Given the differences on Tc –Ta observed in 2010, these datasets
were discarded from the calculation of the CWSI. The NWSBs built
with data from 2009 and 2011 (Table 1) were used as a lower
limit to compute the CWSI (Fig. 8). The CWSI for control trees was
consistently close to 0, except for the experiment of mandarin in
2011; a failure in the system stopped the irrigation for about two
weeks and, as a consequence, the CWSI sharply increased. Once the
irrigation was resumed, it attained normal values close to 0. In 2009,
the stressed trees in both species showed a mean value of about 0.2
at the beginning of the experiment and increased during the stress
period until it reached a maximum value of 0.85 (Fig. 8A and C). In
2011, both species showed a peak between DOY 180 and 190 with a
maximum value near 0.8. Given the high level of water stress expe-
rienced, irrigation was increased in this treatment and that caused
the CWSI values to approach 0. A second period of water deficits
was imposed, and CWSI increased again to a maximum value of
0.8 (Fig. 8B and D). In 2011, the RDI1mandarin and orange trees
were rewatered at the end of the August and mean CWSI values
decreased to values similar to control trees.
The suitability of CWSI for assessing water status in both citrus
species can be observed in its relationship with SWP (Fig. 9), where
data from the two years were pooled together. CWSI showed good
agreement (p < 0.05) with SWP, with R2 yielding 0.66 and 0.59 for
mandarin and orange, respectively. These R2 values were similar for
the orange trees and higher for mandarin than those obtained for
the relationships between Tc –Ta and SWP (Fig. 5). For both species,
the data followed an exponential relationship. A single relationship
was observed for the two treatments, C and RDI1. For CWSI below
0.3 and 0.4 for mandarin and orange respectively, SWP maintained
a constant value of about −1 MPa. Beyond these thresholds of CWSI,
the SWP sharply decreased.
Once the NWSB had been established, a map of CWSI was
derived from the airborne thermal imagery acquired on DOY 217,
2009, showing the spatial distribution of water status for every tree
of the field (Fig. 10). It can be clearly observed that CWSI was related
to the level of applied water, following the treatment distribution.
In fact, while the average CWSI of the control plots was 0.35, the
average CWSI of the RDI1 plots reached 0.66, despite the large vari-
ability encountered in the field. The RDI1 plots that were located
uphill (northern part of the experimental field) were more stressed
as compared to those plots located downhill.
100 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104

7 7
A 2009 D 2009
6 6

5 5

4 4
Tc-Ta

Tc-Ta
3 3

2 2
C C
1 1
RDI RDI1
0 0
170 190 210 230 250 170 190 210 230 250
DOY DOY

7 7
B 2010 E 2010
6 6
5 5
4 4
Tc-Ta

Tc-Ta
3 3
2 2
C C
1 1
RDI1 RDI1
0 0
170 190 210 230 250 170 190 210 230 250
DOY DOY

7 7
C 2011 F 2011
6 6
5 5
4 4
Tc-Ta

Tc-Ta

3 3
2 2
C C
1 1
RDI1 RDI1
0 0
170 190 210 230 250 170 190 210 230 250
DOY DOY

Fig. 6. Time evolution of the difference between canopy and air temperature (Tc –Ta , ◦ C) for mandarin (Fig. 4A–C) and orange (Fig. 4D–F) for 2009, 2010 and 2011 measured
at noon. The bold lines correspond to the five-day moving average for each treatment, C (dashed line) and RDI1 (solid line).

Tc-Ta (ºC) Tc-Ta (ºC)

0 2 4 6 0 2 4 6
0.0 0.0
A C B C
-0.5 RDI1 -0.5 RDI1

-1.0 -1.0
SWP (MPa)
SWP (MPa)

-1.5 -1.5

-2.0 -2.0

-2.5 -2.5 R2=0.55*

R2=0.56**
-3.0 -3.0

-3.5 -3.5

Fig. 7. Relationships between stem water potential (SWP; MPa) and Tc –Ta (◦ C) for mandarin (Fig. 5A) and orange plots (Fig. 5B). Data from 2009 and 2011 were pooled
together. A negative exponential regression line was adjusted in both species. The correlation coefficient is shown in the graph. The level of significance is shown as:
*(p < 0.01), **(p < 0.001).
 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104 101

1.0 1.0
A C C C
0.8 RDI1 0.8 RDI1

0.6 0.6

0.4 0.4

CWSI
CWSI

0.2 0.2

0.0 0.0
170 190 210 230 250 170 19 0 21 0 23 0 25 0
-0.2 -0.2

-0.4 -0.4
DOY DOY
1.0 1.0
B C D C
0.8 0.8 RDI1
RDI1

0.6 0.6

0.4 0.4

CWSI
CWSI

0.2 0.2

0.0 0.0
170 190 210 230 25 0 170 19 0 21 0 23 0 25 0
-0.2 -0.2

-0.4 -0.4
DOY
DOY

Fig. 8. Time evolution of the CWSI for mandarin (A and B) and orange (C and D) for 2009 and 2011. The bold lines correspond with five-day moving average for treatment C
(dashed lines) and RDI1 (solid lines).

4. Discussion
This paper demonstrated that CWSI is a suitable indicator for
the assessment of water status in citrus, although there are some
specific features that deserve special attention, when compared to
its use in annual crops (Moran, 1994) or in other orchard trees (e.g.,
Testi et al., 2008). Robust upper and lower limits are needed to char-
acterize the CWSI with precision (Jackson, 1982). Even though the
linear regressions representing the NWSB (Fig. 3) were always sig-
nificant (p < 0.05), the data had more scattering than that of other
tree crops where the NWSB had been assessed, such as pistachio
(Testi et al., 2008) and sweet lime (Sepaskhah and Kashefipour,
1994). This is probably due to the fluctuations in canopy temper-
ature shown in Fig. 1 that requires careful assessment of the time
interval to estimate an average canopy temperature in relation to
the air temperatures, as it was described in Fig. 2. Fluctuations in
stomatal resistance in citrus have been described long ago (Levy and
Kaufmann, 1976) and more recently, fluctuations in citrus canopy
conductance have been described as well (Dzikiti et al., 2007). These
special features may be at the origin of the difficulties encountered
in the past to assess CWSI in citrus (e.g., Ballester et al., 2013a). We
have also demonstrated that no unique NWSB can be ascribed to

CWSI CWSI
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0.0 0.0
A C B C
-0.5 -0.5
RDI1 RDI1
-1.0 -1.0
SWP (MPa)

-1.5
SWP (Mpa)

-1.5

-2.0 -2.0
R2=0.59**
-2.5 -2.5
R2=0.66**
-3.0 -3.0
-3.5 -3.5
-4.0 -4.0

Fig. 9. Relationship between stem water potential (SWP; MPa) and CWSI for mandarin (Fig. 7A) and orange (Fig. 7B). Data from 2009 and 2011 were pooled together. A
negative exponential regression line was adjusted in both species. The correlation coefficient is shown in the graph. The level of significance is shown as in Fig. 7.
102 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104
Fig. 10. Diagram of the experimental design of the orange experiment and its corresponding calculated CWSI map obtained at 35 cm resolution.
the citrus species, as the line for mandarin was different from that
of orange (Fig. 3). The variability in the response to water deficits
among citrus species was clearly depicted in the early review of
Kriedemann and Barrs (1981) and more recently, research dealing
with the optimization of deficit irrigation strategies has described
the contrasting behavior of mandarin (cv. Clementina de Nules) and
orange (cv. Navel Lane Late) in response to RDI strategies (Ballester
et al., 2011, 2013b). Finally, we have also found that crop load has
an influence on the NWSB (Fig. 7), although it appears that it is the
absence of fruits that have the most effect on the intercept of the
NWSB (Fig. 7). This may be a general phenomenon in other species,
given the source–sink interactions in fruit trees and their effects
on stomatal and canopy conductances (DeJong, 1986; Reyes et al.,
2006), and therefore on canopy temperature.
CWSI would be considered for irrigation management purposes,
as changes in the crop water status were translated into a propor-
tional shift of the index (Fig. 9). When CWSI is compared to Tc –Ta
as a proxy of water status (by comparing the R2 and p value of
the relationship of both indices to SWP, Figs. 5 and 9), it can be
observed that CWSI represents more accurately the crop water sta-
tus, especially for mandarin. Compared to Tc –Ta , the performance
of the CWSI is improved by the consideration of the evaporative
demand, as well as the species-specific response of water relations
to VPD.
The NWSB of citrus found here is similar to that reported for
olive (Berni et al., 2009), a tree crop well adapted to Mediterranean-
type climatic conditions. In both species, Tc is generally higher than
Ta for the usual values of VPD and their NWSB display a steeper
slope compared with other tree crops, such as pistachio (Testi et al.,
2008). As far as we know, the only NWSB that has been published in
citrus was that presented by Sepaskhah and Kashefipour (1994) in
sweet lime. Canopy temperature was measured with a hand-held
infrared thermometer 2 m away from the tree, around 15◦ above
the horizontal, a procedure that would not assess the Tc of the top
of the canopy. This may be the reason why the values obtained
by them were considerably different from what we have observed.
In their case, Tc was lower than Ta for well watered conditions,
resulting in negative values of Tc –Ta for a wide range of VPD. In
our case, these values were always positive. Moreover, the same
authors demonstrated that water use of sweet lime is quite differ-
ent from other citrus (Sepaskhah and Kashefipour, 1995); it could
also explain the differences encountered in the two studies. Other
studies report that transpiration in citrus is less than that of other
fruit tree species during summer, also under well watered condi-
tions (Villalobos et al., 2013). The low transpiration rate agrees with
a partial stomatal closure that results in an increase of canopy tem-
perature. This characteristic of citrus grove is well known and has
been characterized in the past (Veste et al., 2000; Villalobos et al.,
2009).
The different behavior observed in Tc –Ta during 2010 as com-
pared to the other two years was attributed above to the lack of fruit
production in that year. The occurrence of vegetative flush growth
that is enhanced in off-years (Lenz, 1967) could be at the origin of
this behavior. In citrus, these flushes of growth take place from two
to four times, depending on cultivar and climate (Monselise and
Goldschmidt, 1982), and occur in spring, summer and autumn. We
have shown (Table 2) that recently-formed leaves displayed a con-
trasted gas exchange as compared to over-wintered or older leaves.
Syvertsen et al. (1981) found that the transpiration rate of sum-
mer leaves were 25% higher compared to spring leaves. Stuckens
et al. (2011) obtained similar results in a study where canopy tem-
perature was monitored during three seasons. The new shoots are
mainly located in the upper part of the canopy, well within the
field of view of the IRT sensors positioned above the tree crowns
in a zenital view. In addition to our observations of Table 2, two
different datasets supported the hypothesis that the summer new
growth would have an effect on the crown temperature measured
by the IRT sensors. The information acquired with the hyperspec-
tral imager that was flown over the orchard in September 2010
and 2011 showed that the top of the crowns had a lower amount of
chlorophyll in the low-crop year (2010) as compared to a normal
year. We hypothesized that this difference was related to the newly
formed leaves that had lower chlorophyll content than the older
leaves. A possible effect of nutrient status on chlorophyll content
would be discarded, as fertilizer application was managed similarly
among years and treatments, and was never limiting. Furthermore,
no pest or disease was detected during the experiment that would
damage the canopy. The relationship between crop load and the
intercept of the NWSB for each year also pointed out to a more
extended vegetative growth in 2010. The lower the crop load, the
higher the vegetative growth, and according to the results obtained
here and elsewhere (Syvertsen et al., 1981) the higher the stomatal
conductance. Spiegel-Roy and Goldschmidt, 1996 stated that sto-
matal regulation of water loss in the young leaves of citrus is not
well developed. The effect of growth habit on the assessment of
crop water status based on thermal information appears impor-
tant for the quantification of indices such as the CWSI in citrus.
Research is needed to investigate whether this effect may affect
other evergreen species.
One of the shortcomings for the use of the CWSI in years that
would have substantial vegetative growth in the summer due to a
very low crop load is the lack of representativeness of control trees
needed to develop the NWSB. Contrary to the case of olive, modern
citrus cultivars have relatively low alternate bearing behavior and
the situation we observe in 2010 of very low crop load is not com-
mon. Therefore, the CWSI approach is valid as an indicator of tree
water deficits, given its correlation with established methods of
measuring tree water status. Further research is needed to develop
 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104
a methodology that integrates leaf age and enables the monitoring
of the water status when vegetative flushes occur.
Knowledge of the natural variability generally observed in field
(which is related to either soil and/or plant heterogeneity or to
the lack of uniformity in irrigation water distribution) is essential
for the correct application of deficit irrigation (DI). The commer-
cial development of DI strategies requires the tree water status
to be maintained within a narrow range in order to save water
without reducing yield. In this context, high-resolution thermal
imagery proposed in this work would be an ideal tool, as it enables
the monitoring of large areas with the precision required to accu-
rately manage DI. The robustness of this approach has already been
demonstrated in several fruit tree species (Berni et al., 2009; Meron
et al., 2010; Gonzalez-Dugo et al., 2013). The remote assessment of
water status at the tree level for whole fields enables the economic
analysis and risk evaluation of water application under different
strategies (Gonzalez-Dugo et al., 2014). The large range of variabil-
ity under plots that were similarly managed can be derived from
the thermal image obtained over this experiment. This information
could be used to monitor irrigation by a periodic assessment of field
water status (regular flights), by establishing control points that
characterize the variability of the orchard or, in high-value crops,
by assessing the most sensitive area in order to monitor those trees
displaying the lowest water status.
5. Conclusions
It can be concluded that CWSI is a suitable indicator for water
status monitoring in citrus during summertime in semi-arid con-
ditions, and also for manage irrigation in precision agriculture.
Nevertheless, some specific considerations must be taken into
account if the CWSI is to be used in citrus, namely, the short-term
oscillations of canopy temperature and the vegetative flushes. The
effect of the short-term oscillations could be avoided by adapting
the interval considered for averaging thermal data to the oscilla-
tions. The flushes of new growth that occur mainly in spring and
summer alter the water relation of the top view canopy and shift the
NWSB. The main shortcoming is the inability of well-irrigated trees
to serve as a control, as the development of these flushes would
differ among trees with a contrasted water status. Further research
will characterize this behavior and will propose adaptations to take
leaf age into account.
In this work, the NWSB was obtained from point IRT sensors and
the methods scaled up to high resolution aerial thermal imagery to
obtain CWSI maps at 35 cm resolution. The utilities of such tool
increases, as the water status of a whole field can be assessed and
thus, studies related with spatial variability can be tackled.
Acknowledgements
Authors acknowledge K. Gutierrez, M. Medina, D. Notario, R.
Romero and A. Vera for their technical support. Tepro S.L. is also
acknowledged. This work was funded by the Spanish Ministry of
Science and Innovation for the projects CONSOLIDER CSD2006-
0067 and AGL2012-40053-C03-01, with participation of funds from
FEDER (European Union).
References
Allen, R.G., Pereira, J.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration: Guide-
lines for Computing Crop Water Requirements. FAO Irrigation and Drainage
Paper, vol. 56. Food and Agriculture Organization of the United Nations, Rome,
pp. 300.
Ballester, C., Castel, J., Intrigliolo, D.S., Castel, J.R., 2011. Response of Clementina
de Nules citrus trees to summer deficit irrigation. Yield components and fruit
composition. Agric. Water Manag. 98 (6), 1027–1032.
103
Ballester, C., Jimenez-Bello, M.A., Castel, J.R., Intrigliolo, D.S., 2013a. Usefulness of
thermography for plant water stress detection in citrus and persimmon trees.
Agric. For. Meteorol. 168, 120–129.
Ballester, C., Castel, J., Intrigliolo, D.S., Castel, J.R., 2013b. Response of Navel Lane Late
citrus trees to regulated deficit irrigation: yield components and fruit composi-
tion. Irrig. Sci. 31 (3), 333–341.
Bastiaanssen, W.G.M., Bos, M.G., 1999. Irrigation performance indicators based on
remotely sensed data: a review of literature. Irrig. Drain. Syst. 13, 291–311.
Ben-Gal, A., Agam, N., Alchanatis, V., Cohen, Y., Yermiyahu, U., Zipori, I., Presnov, E.,
Sprinstin, M., Dag, A., 2009. Evaluating water stress in irrigated olives: correla-
tion of soil water status, tree water status, and thermal imagery. Irrig. Sci. 27 (5),
367–376.
Berni, J.A.J., Zarco-Tejada, P.J., Sepulcre-Canto, G., Fereres, E., Villalobos, F., 2009.
Mapping canopy conductance and CWSI in olive orchards using high resolution
thermal remote sensing imagery. Remote Sens. Environ. 113, 2380–2388.
DeJong, T.M., 1986. Fruit effects on photosynthesis in Prunus persica. Physiol. Plant.
66, 149–153.
Dzikiti, S., Steppe, K., Lemeur, R., Milford, J.R., 2007. Whole-tree level water balance
and its implications on stomatal oscillations in orange trees [Citrus sinensis (L.)
Osbeck] under natural climatic conditions. J. Exp. Bot. 58 (7), 1893–1901.
FAOSTAT, 2012. Statistical Databases. FAO, Food and Agriculture Organization of the
United Nations, http://faostat.fao.org/. Data retrieved in 2013.
Fereres, E., Goldhamer, D.A., Sadras, V.O., 2012. Yield response to water of fruit trees
and vines: guidelines. In: Steduto, P., Hsiao, T.C., Fereres, E., Raes, D. (Eds.), Crop
Yield Response to Water. FAO Irrigation and Drainage Paper, vol. 66. Food and
Agriculture Organization of the United Nations, Rome, pp. 246–296.
Fernandez, J.E., Cuevas, M.V., 2010. Irrigation scheduling from stem diameter varia-
tions: a review. Agric. For. Meteorol. 150, 135–151.
Gates, D., 1964. Leaf temperature and transpiration. Agron. J. 56, 273–277.
Goldhamer, D.A., 2012. Citrus. In: Steduto, P., Hsiao, T.C., Fereres, E., Raes, D. (Eds.),
Crop Yield Response to Water. FAO Irrigation and Drainage Paper, vol. 66. Food
and Agriculture Organization of the United Nations, Rome, pp. 316–331.
Gonzalez-Dugo, V., Zarco-Tejada, P.J., Berni, J.A.J., Suarez, L., Goldhamer, D.A., Fereres,
E., 2012. Almond tree canopy temperature reveals intra-crown variability that
is water stress-dependent. Agric. For. Meteorol. 154–155, 156–165.
Gonzalez-Dugo, V., Zarco-Tejada, P.J., Nicolas, E., Nortes, P.A., Alarcón, D.S., Fereres,
E., 2013. Using high resolution UAV thermal imagery to assess the variability in
the water status of five fruit tree species within a commercial orchard. Precis.
Agric. 14 (6), 660–678.
Gonzalez-Dugo, V., Goldhamer, D.A., Zarco-Tejada, P.J., Fereres, E., 2014. Improv-
ing the precision of irrigation in a pistachio farm using an unmanned airborne
thermal system. Irrig. Sci. (in press).
González-Dugo, M.P., Moran, M.S., Mateos, L., Bryant, R., 2006. Canopy temperature
variability as an indicator of crop water stress severity. Irrig. Sci. 24 (4), 233–240.
Hsiao, T.C., 1973. Plant responses to water stress. Ann. Rev. Plant Physiol. 24,
519–570.
Idso, S.B., Jackson, R.D., Reginato, R.J., 1978. Extending the degree day concept of
plant phenological development to include water stress effects. Ecology 59,
431–433.
Idso, S.B., Jackson, R.D., Pinter, P.J.J., Reginato, R.J., Hatfield, J.L., 1981. Normalizing
the stressdegree-day parameter for environmental variability. Agric. Meteorol.
24, 45–55.
Irmak, S., Haman, D.Z., Bastug, R., 2000. Determination of crop water stress index
for irrigation timing and yield estimation of corn. Agron. J. 92 (6), 1221–1227.
Jackson, R., 1982. Canopy temeperature and crop water stress. Adv. Irrig. 1, 43–85.
Jackson, R., Reginato, R., Idso, S.B., 1977. Wheat canopy temperature: a practical tool
for evaluating water requirements. Water Resour. Res. 13 (3), 651–656.
Jackson, R., Idso, S., Reginato, R., Pinter, P.J., 1981. Canopy temperature as a crop
water stress indicator. Water Resour. Res. 17, 1133–1138.
Jifon, J.L., Syvertsen, J.P., Whaley, E., 2005. Growth environment and leaf anatomy
affect nondestructive estimates of chlorophyll and nitrogen in Citrus sp. leaves.
J. Am. Soc. Hortic. Sci. 130 (2), 152–158.
Jones, H.G., 1999. Use of infrared thermometry for estimation of stomatal conduc-
tance as a possible aid to irrigation scheduling. Agric. For. Meteorol. 95, 139–149.
Kar, G., Kumar, A., 2007. Surface energy fluxes and crop water stress index in ground-
nut under irrigated ecosystem. Agric. For. Meteorol. 146, 94–106.
Kriedemann, P.E., Barrs, H.D., 1981. Citrus orchard. In: Kozlowski, T.T. (Ed.), Water
Deficits and Plant Growth. Woody Plant Communities, vol. 6. Academic Press,
New York.
Lenz, F., 1967. Relationships between the vegetative and reproductive growth of
Washington navel orange cuttings (Citrus sinensis L. Osbeck). J. Hortic. Sci. 42,
31–39.
Levy, Y., Kaufmann, M.R., 1976. Cycling of leaf conductance in citrus exposed to
natural and controlled environments. Can. J. Bot. 54, 2215–2218.
Maes, W.H., Steppe, K., 2012. Estimating evapotranspiration and drought stress with
ground-based thermal remote sensing in agriculture: a review. J. Exp. Bot. 63
(13), 4671–4712.
Meron, M., Tsipris, J., Orlov, V., Alchanatis, V., Cohen, Y., 2010. Crop water stress
mapping for site-specific irrigation by thermal imagery and artificial reference
surfaces. Precis. Agric. 11 (2), 148–162.
Monselise, S.P., Goldschmidt, E.E., 1982. Alternate bearing in fruit trees. Hortic. Rev.
4, 128–173.
Moran, M.S., 1994. Irrigation management in Arizona using satellites and airplanes.
Irrig. Sci. 15, 35–44.
Naor, A., 2006. Irrigation scheduling and evaluation of tree water status in deciduous
orchards. Hortic. Rev. 32, 111–165.
104 V. Gonzalez-Dugo et al. / Agricultural and Forest Meteorology 198–199 (2014) 94–104
Rana, G., Katerji, N., de Lorenzi, F., 2005. Measurement and modelling of evapotrans-
piration of irrigated citrus orchard under Mediterranan conditions. Agric. For.
Meteorol. 128, 199–209.
Reyes, V.M., Girona, J., Marsal, J., 2006. Effect of Late Spring defruiting on net
CO2 exchange and leaf area development in apple tree canopies. J. Hortic. Sci.
Biotechnol. 81 (4), 575–582.
Sepaskhah, A.R., Kashefipour, S.M., 1994. Relationships between leaf water potential,
cwsi, yield and fruit-quality of sweet lime under drip irrigation. Agric. Water
Manag. 25 (1), 13–22.
Sepaskhah, A.R., Kashefipour, S.M., 1995. Evapotranspiration and crop coefficient
of sweet lime under drip irrigation. Agric. Water Manag. 27 (3–4), 331–
340.
Sepulcre-Cantó, G., Zarco-Tejada, P.J., Jimenez-Muñoz, J.C., Sobrino, J.A., de Miguel, E.,
Villalobos, F.J., 2006. Detection of water stress in an olive orchard with thermal
remote sensing imagery. Agric. For. Meteorol. 136 (1–2), 31–44.
Spiegel-Roy, P., Goldschmidt, E.E., 1996. Biology of Citrus, vol. 229. Cambridge Uni-
versity Press, Cambridge, UK.
Spreen, T.H., 2001. Projections of world production and consumption of citrus to
2010. In: Ministry of Agriculture and the FAO (Ed.), Proceedings of the Inter-
national Citrus Symposium. International Citrus Symposium, Beijing, China, pp.
152–164.
Stuckens, J., Dzikiti, S., Verstraeten, W.W., Verreynne, S., Swennen, R., Coppin, P.,
2011. Physiological interpretation of a hyperspectral time series in a citrus
orchard. Agric. For. Meteorol. 151 (7), 1002–1015.
Syvertsen, J.P., Smith, M.L., Allen, J.C., 1981. Growth rate and water relations of citrus
leaf flushes. Ann. Bot. 47, 97–105.
Testi, L., Goldhamer, D., Iniesta, F., Salinas, M., 2008. Crop water stress index is a
sensitive water stress indicator in pistachio trees. Irrig. Sci. 26, 395–405.
Turrell, F.M., 1961. Growth of the photosynthetic area of citrus. Bot. Gaz. 122,
284–298.
Veste, M., Ben-Gal, A., Shani, U., 2000. Impact of thermal stress and high VPD on gas
exchange and chlorophyll fluorescence of citrus grandis under desert conditions.
Acta Hortic. 531, 143–149.
Villalobos, F.J., Testi, L., Moreno-Perez, M.F., 2009. Evaporation and canopy conduc-
tance of citrus orchards. Agric. Water Manag. 96 (4), 565–573.
Villalobos, F.J., Testi, L., Orgaz, F., Garcia-Tejera, O., Lopez-Bernal, A., Gonzalez-
Dugo, V., Ballester-Lurbe, C., Castel, J.R., Alarcon-Cabañero, J.J., Nicolas-Nicolas,
E., Girona, J., Marsal, J., Fereres, E., 2013. Modelling canopy conductance and
transpiration of fruit trees in Mediterranean areas: a simplified approach. Agric.
For. Meteorol. 171–172, 93–103.
Wang, D., Gartung, J., 2010. Infrared canopy temperature of early-ripening peach
trees under postharvest deficit irrigation. Agric. Water Manag. 97 (11),
1787–1794.
Wanjura, D.F., Hatfield, J.L., Upchurch, D.R., 1990. Crop water stress index relation-
ships with crop productivity. Irrig. Sci. 11 (2), 93–99.
Zarco-Tejada, P.J., Miller, J.R., Morales, A., Berjón, A., Agüera, J., 2004. Hyperspectral
indices and model simulation for chlorophyll estimation in open-canopy tree
crops. Remote Sens. Environ. 90 (4), 463–476.
Zarco-Tejada, P.J., Gonzalez-Dugo, V., Berni, J.A.J., 2012. Fluorescence, temperature
and narrow-band indices acquired from a UAV platform for water stress detec-
tion using a micro-hyperspectral imager and a thermal camera. Remote Sens.
Environ. 117, 322–337.